Hospital Volume and Mortality After Transjugular Intrahepatic

Home , Liver transplantation

Page 1of28 Authors: 4 3 2 1 Title: Center/HarvardMedical MA;Boston,School, AS: Assi Professor Key Words:Key Professor Center/Harvardmedical MA;Boston,School, Asso MC: MuneebAhmed Short Title: Short States the Negev,BeerSheva, the VN:Israel; Professorand He DivisionGastroenterology, of Department ofMedici DivisionGastroenterology of Hepatology,and Depar Clinical Research Center,Clinical Research Soroka University Medical Vascular andof Division Interventional Radiology, between this versionandtheVersionofrecord. Pleasecitethis articleasdoi:10.1002/hep.29354. through thecopyediting, typesetting, paginationandproofreadingprocess, whichmayleadtodifferences This istheauthormanuscript acceptedforpublicationandhasundergone full peerreviewbuthasnotbeen

Hospitalvolume and mortalityafter transjugular i

AmmarSarwar Effectofhospital volumeTIPS on mortality volumeoutcome relationship;mortality;admi TIPS; 1

Author Manuscript

1 , Lujia ZhouLujia ,

DepartmentofRadiology,Beth IsraelDeaconess Medi ne, University HealthMichiganne, of System;EBT: Assi ad ofDepartmentad Center and FacultyCenter Health, BenGurion of Universit tmentofMedicine, Deaconess Beth MedicalIsrael 2 stant Professor,stant LZ:Research Fellow; Associate MA: ,Elliot B. Tapper ntrahepatic portosystemicntrahepatic shuntcreationin theUni ciateProfessor, RM: Professor Assistant nistrativedata 3,4 , , MichaelCurry

3 Raza MalikRaza 3 , , stant y of yof ted ted cal

APRDRGs: All PatientAPRDRGs: RefinedDiagnosis Related Gro Analysis VarianceANOVA: of Agency AHRQ: Research Healthcare andfor Quality Abbreviations: Ammar Sarwar, Ammar M.D. Disclosures: Support: Grant intrahepaticTIPS: Transjugular portosystemic shunt State SID: Inpatient Database SpontaneousSBP: bacterial peritonitis NationwideNRD: Database Readmission Nationwide NIS: Sample Inpatient NAFLD:nonalcoholic fattyliver disease EndstageMELD:Model Liver Diseaseof ICU: IntensiveCare Unit InternationalICD9: Classificationof Diseasesver hepatorenalHRS: syndrome hepaticHE: encephalopathy HCUP:Healthcare Cost Utilizationand Project EVAR:EndovascularRepair Aortic Email: Email: 6177542523Phone: Fax: 6177542545 Road,Deaconess 1Boston, MA02215 IsraelDeaconess Beth Medical Center DepartmentofRadiology, 308B WCC Correspondence [email protected] None ofhavetheauthors conflicts interest rel of

None.

Author Manuscript:

Page 3of28 The link between higher procedure volume and better and volume procedure higher between link The ABSTRACT e ie t dtrie f rcdr vlm affected volume procedure if determine to aimed We assessed for hospitalsfor assessed o vr hg vlm hsias p00) Eetv a Elective p<0.01). hospitals; volume high very for o optl promn ≥0 IS e ya, ny hos only year, per TIPS ≥30 performing hospitals to annual TIPS volume (20.3% for very low to 30.8% for 30.8% to low very for (20.3% volume TIPS annual decreased Mortality female). 37.5% years; (57±10.9 nreai proytmc hn (TIPS). shunt portosystemic intrahepatic 9/year), medium (1019/year), high (2029/year), an (2029/year), high (1019/year), medium 9/year), p=0.01), 59/year (aOR: 2.0, 95%CI:1.253.17; p<0.0 95%CI:1.253.17; 2.0, (aOR: 59/year p=0.01), optlztos uig 03 n h Uie Sae ( States United the in 2013 during hospitalizations patients’ ≥18 years old undergoing TIPS during a ho a during TIPS undergoing old years ≥18 patients’ between riskadjusted mortality rate for very low v low very for rate mortality riskadjusted between 1. [aOR: (2029/year mortality inpatient higher had rnpatto wr selected. were transplantation .%. IS oue f 2 TP/er vrca bleed variceal TIPS/year, ≤20 of volume TIPS 7.8%). higherfactors for mortality benefitsofpatienand inpatient mortality is lower in hospitals performin hospitals in lower is mortality inpatient factors for inpatient mortality in patients with bo with patients mortalityinpatientin for factors

Author Manuscript

categorized into quintiles based on annual TIPS vol TIPS annual on based quintiles into categorized

n pdmooia aayi o a alpyr databas allpayer an of analysis epidemiological An th elective and emergent admissions.emergent andelective th t tohighert transfer volumeis centers warranted. g ≥20 TIPS per year. Future research exploring prev exploring research Future year. per TIPS ≥20 g Hepatology Hepatology olume and very high volume hospitals was 6.1% (13.9 6.1% was hospitals volume high very and olume d very high (≥30/year). TIPS were placed in 5529 pa 5529 in placed were TIPS (≥30/year). high very d ainie edisos aaae ws efre. Al performed. was Database) Readmissions Nationwide spital admission (n=5529) without concurrent or pri or concurrent without (n=5529) admission spital msin wr mr cmo i hsias ih higher with hospitals in common more were dmissions with rising annual TIPS volume (13% for very low to low very for (13% volume TIPS annual rising with 1), and 1019/year (aOR: 1.9, 95%CI:1.173.00; p=0. 95%CI:1.173.00; 1.9, (aOR: 1019/year and 1), , 5C:.428; =.9) Te boue differen absolute The p=0.19]). 95%CI:0.842.84; 4, ias efrig /er aR 19 95%CI:1.213. 1.9, (aOR: 14/year performing pitals very high; p<0.01). On multivariate analysis, comp analysis, multivariate On p<0.01). high; very otoe o ugclpoeue swl establish well is procedures surgical for outcomes n, n nscma ifcin wr idpnet ris independent were infections nosocomial and ing, nain mraiy n ains negig transjug undergoing patients in mortality inpatient ume; very low (14/year), low (5 low (14/year), low very ume; Conclusions: utd otlt was mortality justed recording e The risk of risk The entable or liver or % vs. % tients ared 6% 6% ular 01) ed. 01; ce 3 k k l

demonstrated most for majorprocedures.(1–surgical betweenassociationThe withhospitals annua higher also been reportedbeenalso percutaneousfor cardiovascular hospital and volumehospitaland characteristics explaimaythat volumeinpatientand mortality imageguided profor Given the variationsknown in mortalitypatient for volumeinpatientTIPS on outcomes likely is warrant wereconditionslikely more undergoplaceme to TIPS peripheral arterialperipheralinterventions, and endovascular centers byprocesscenters a knownregionalization as leads theseon severalfindings, have studies shown that variationinpatientin mortalitypatientswith for account variationsaccount for observed inpatientin mortal hospitalvolume ofadmissions patients withfor cir interventionalbecan procedurescomplex, associate notyetbeenhasstudy expanded percutaneous to pro admissions withadmissionsplacement TIPS 12.5% from 2003in t ofstudyNationwiderecent the SampleInpatient (NI intrahepaticTransjugular portosystemic shunt(TIPS criticallyillpatients. requires a highdegreerequires of andtechnical clinical ex

Author Manuscript

This article isprotected by copyright. All rights reserved. cirrhosis can becirrhosiscan a resultofhospitallevel factors s withand s cirrhosis relationshipbetween annualpr perience toachieveperience optimal patientoutcomes.(15,16 rhosis oresophageal varicealrhosis alonedoesbleeding n ity, although patientsatalthoughity,volume hospitalshigh wit Hepatology Hepatology abdominalaorticaneurysm repair (EVAR).(5–10)Bui n variationsnmortality inprocedures.after TIPS increasingreferralofelective EVARhigh cases to cedures, wecedures, sought preprocedure to identify patien S) aS) reported reduction inpatientin mortalityduri ed.(18,19) ) placement treating) for sequalaeofportal hyperte interventions,suchas coronaryvalvular and proced toimproved patientoutcomes.(11–14) However,such d withdoperator an learning curve,are performeand l procedure volumeprocedurel and improvedsurvival been has 4) Better patient4) Better athigher outcomes volume centers nt, suggestingnt, additionalthat influenonstudy the ceduresnoncardiovascularin wherepatients, o2012.(17)in10.6% Studies have demonstrated that .(18) .(18) annualThe h theseh ng ocedure volume ot )A nsion ce of ce t, lding ures, din Page 4of28 have

Page 5of28 StudyProcedurepopulation: diagnosiscodes allfor oversight board becauseprotected nohealth informa abilitycontrol multipleto for hospitalizations wi allhospitalizationseach patientallowing for anal generalizability,wereweights dischargedeveloped Health Information SystemsInformationHealth adjust to casemixfor s RefinedDiagnosis Related Groups (APRDRGs) severit principal to the diagnosislikelyand have origi to smaller HCUP smaller database sample,7.8(20% annu million Elixhausercomorbidity measures areassignedthat validated as discriminative the most predictiveand comorbidity(AHRQ) software.(21) These idemeasures diagnoses,discharge inpatient procedures, length o 49.1%representing ofallUS hospitalizations. It r is rare, patients rare, iswith multiplecodes of39.1ICD9 39.1 (Intraabdominalvenous shunt),and age ≥18ye The study wasstudyThe reviewed institutionalby the review weightsdischarge priortoanalysis. were stratified, bywere patientstratified, hospital and characteri procedures. procedures. criteriaInclusion were International C Utilization includesProject that(HCUP) allhospit Source: Data NationalThe ReadmissionDatabase (NRD METHODS This allpayer This (insuredand uninsured)database con outcomes patients,cirrhoticfor therefore patients

Author Manuscript

This article isprotected by copyright. All rights reserved. th a TIPS procedure a code TIPS same patient.th the for e To nated priorhospital nated to the It includesalsostay. ysis at a patientata ysis rather thandischarge level p and stics. Nationalstics.estimates were calculated by applyi ecordsdeidentified patient and hospitaldemograph alizations2013during in 21 geographicallydiverse ICD9withprocedure code(liver50.5*transplanta were (n=58) excluded. Liver changetransplantation lassificationofDiseases 9 Hepatology Hepatology mortalityrisk score for patientscirrhoticin the board as appropriate as exemption board institutionfor from hospitalizations theindatabase were searched for in theinNRD to produce nationalestimatesafter the everity.APRDRG The risk ofmortality score has be f stay,dischargeand status. f Inaddition,inclu it tains dataapproximately tains from million14 discharges usingtheAgency Healthcarefor Research and Qualit tion wastion available data. the in ars. Since placementofars. multiplesingle ap in TIPS ntify coexistingntify medicalnotconditionsdirectly r al discharges).(22)al contrast In totheNIS,the NRD yassignedmeasures usingsoftware developed by 3M ) isdatabase a ) developed Costan Healthcare by the th version (ICD9) procedureversion (ICD9) code All Patient NIS,a roviding the nsure nsure des29 ng ng states.(20) tion) tion) ics, ics, al reviewal s data TIPS elated , atient en links y 5 d

APRDRG riskAPRDRGofmortality of(likelihooddying: min (n=87) ordiagnosis (n=87) ICD9 996.82code (complication diagnosis codes diagnosis ofcirculatory shockaand previous orabsence presence ofintensive unit (ICU)care ad the presence the ofmedical candiagnosesthat cause (o typicallycirrhoticin patientsfound ICU (Appendix disease, viraldisease, livernonalcoholic disease, l fatty important predictor important of mortalityafterbut TIPS, th Patients were categorizedPatients by age, insurancegender, hospitalizations prior excluded.were and procedure codesprocedureand orforhydrothorax, ascites hep validatedinstrumentsChildPughsuch as orModel f features ofhepatic decompensationfeatures were controlled Hospitalscovariates included ownership (governmen acidosis],hypokalemia, substanceabuse)and (Appen renal (acute failure,infection,diabetes, coagulop transplantation)hospitalizationsduring ent the in with area <1million residentsand micropolitan are Indexhospitalizations categorizedwere by mode of nonmetropolitan),locationand by metropolitan siz sizeowned),bed (small, large), medium, teaching s of 28comorbidity AHRQof measures (comorbidity measu bacterial peritonitis,bacterial and variceal(Appen bleeding

Author Manuscript

This article isprotected by copyright. All rights reserved. iver disease iver [NAFLD],other). Severityliver disof e oflaboratoryabsence dataprecluded assessment o ire databasewereire used toidentify transplant hospi athy,respiratory electrolyteimbalance failure, [h Table 1).(23) Hospitalizations1).(23) Table werecharac further dix Table 1). Comorbiditiesdix1). Table bywere assessed thep Hepatology Hepatology as). Patientswithas). ICD9codeprocedure 50.* (liver or, moderate,or, extreme).major, e (metropolitan areawith (metropolitan e >1million residents, met mission. admissionmission. ICU wasusing assessed ICD9 lyclassificationdescribedprocedure of ICD9 code tatus (metropolitan teaching, nonteacmetropolitan admission(electiveemergent), length vs. ofstay a orEndStageLiver Disease(MELD) scores. Therefore atic encephalopathy, atic hepatorenal syndrome, spontane byusingthepresence orabsence ofICD9CM diagno status,and etiologyliver diseaseof (alcoholicl r prolongedr result from) orcomplicated hospitaliza s of liver transplantation)of s either inin (n=39) the t, nonfederal;t, nonprofit;private, private, inves dix1). Table re for liverre diseasefor was excludedanalysis) from an ease is an is ease yponatremia, tals. terizedby dex or iver tor

ropolitan resence f f nd s s hing, tions tions Page 6of28 d , ous sis 6

Page 7of28 r hospital Transplant status was highlycorrelated wi the the multivariateand covariatesmodel, with an Pearsonregression. correlation wastest used for t acrossnonsignificant bivariate and multivariate a it thathavemay onimpact an mortality when contro analysis.multivariate Age was includednon despite generalizedlogisticlinearmixed model regress for variablescategorical and ANOVA continuousfor vari exposureofTheinterest wasproceduralannualTIPS Statistical analysis: Statistical Descriptive werestatistics o nonalcoholicetiologyof disease,liver infection, on clinicalon statisticaland a significance, factors categoriesofannualhospital volumeTIPS with simi 29/year; n=394) very29/year; and high(≥30/year;n=497). Th visualvery estimation); low n=589), low(14/year; admissions,twoseparate generalizedlinearmixed m differencesto Duesignificant observedin mortalit average using patientcharacteristics bybacktrans goodnessoffit statisticsgoodnessoffit wereeach calculated for andTIPS/year), patientincome by quartile zip code sample sizesample ofthetwo subcohorts, volume of TIPS as priorstudies as showing effecton an mortalitywit (including dummy(including variables) withcstatistic a of0 admissions haveadmissionsvariables, 9 and cstatisticsof0. range range reported modelsfor investigating volumeoutc =0.788),so only volumeprocedure was included t in

Author Manuscript

This article isprotected by copyright. All rights reserved. ssociatedwith inpatienthigher mortalitywere sele r btainedusingPearson’s chisquare dichotomousfor andwere diabetes included pvalgivensignificant wereabove considered0.6 to have highinteraction .776; the models electivefor emerge andadmissions h inclusion h of thesevariables. Hospitallevel pred y rate between yrate patients electivewith emergentand model. Forour model. model,full weretherevar 16 total nalysesthuswereand excluded the from multi final 895 and895 0.732, respectively.cstatisticwasOur wi (59/year; n=606), medium(59/year; (1019/year; n=502), hig Hepatology Hepatology ion in order tocorrectionorderin thewithinhospital correla th hospital procedureth TIPS volume (Pearsoncorrelat esting multicollinearitybetweenthetwo any co of forming predicted mortalityforming the from model.final was excludedthemodels.Discrimination from and significant pvalues significant conceptualupon basedhypothe lling predictors. Varicealother for bleeding,alco lar numberlar eachinofprocedures category (assessed was regrouped intowas regrouped two(119 andgroups TIPS/year ≥ omerelationships procedures.(24,25)surgicalfor A eprimaryofinterest outcome was inpatient mortali ables. Multivariateables. analysiswas performedusing he model.final Adjusted mortalitywererates calcu odelswereto thesesubcohorts. Consideringfit th volumeeach hospital. Patientsfor were divided in cted for tions. Basedtions. ues as well as ues ictors wereictors iables holicvs. thin the the thin s. s. variatesin or variable h(20 nt ion ty. lated ll e to to ses ses by 20 7

and analysesand were performedusingSAS (Version9.4 Institute andInstitute Inc.), SPSS (IBMReleased 2015. Corp. NY: IBMsoftware.NY: Corp.) statistical tests werestatistical 2tailed,tests pvalue a and0 of

Author Manuscript

This article isprotected by copyright. All rights reserved. .05was statisticallyconsidered Data significant. Hepatology Hepatology IBM SPSSfor23.0. Statistics Version Arm Windows, oftheSASSystem Copyright 2016for SAS© Windows. management Page 8of28 onk, 8

Page 9of28 (Table 2). Compared2). (Table to performinghospitals ≥30 TIP lower annuallowerhospitalprocedure volume groups 95%CI:1.173.00;p=0.01) hadodds higher ofinpatie 95%CI:1.213.01;p=0.01), 59/year(aOR:95%CI 2.0, p=0.19]). Similarly,p=0.19]). theadjusted mortalityofrate Mode of admission: Modeof mortalityhighest rate the for vs. quintile (13.9% chronic conditionschronic ofcoagulopathy and renal failur Multivariateanalysisrevealed increasing that age, inpatient mortality(Appendix 2). table elective admissions, the unadjusted mortalityrate inpatient mortalitydecreasedratewith increasing admission. Inpatient mortality this wascohort for adult5529Overall, patients underwent without TIPS RESULTS income quartiles patient’sfor ZIPcode, variceal b with ≥20with hospitals)(7%TIPS/year and of thelowest characteristicsofpatients, hospitalizations, and 14 14 TIPS/year,95/443(21%) withhospitals 59 TIPS Medicaidselfpay, alcoholicor etiologyofliver d univariate analysis,In lower volume,increaseTIPS hospitals. (4%) hospitals (4%) with 2029 and 13/443TIPS/year, (3%

Differentiatingpatients withelective an emerg vs. Author Manuscript

7.8%;1).Figure isease (vs nonalcoholic,(vsisease viral,other), hou median hospitalseach for quintile inare reported Table1 thelowesthospitals quintile was nearly twice the wereindependently associated with oddshigher of m 10.5% (583/5529).10.5% Therewere 278/443(63%) hospital leeding, diabetes, leeding, infectionand were associated wi annual volumeTIPS both oftypesadmissions.for Fo Hepatology Hepatology variceal bleeding,and/or infection during theadm for the lowest the for of quintile volumeannualTIPS (14 e, higher incomee,higher quartilespatient ZIP for code, an d patientage,dnonelective primary admission, paye inpatient exclusivelymortalitywere metropolitan priororconcurrent liver duringtransplantation t /year, 39/443 hospitals/year, (9%) with 1019 TIPS/year, nt nt mortality(2029/year1.4, [aOR: 95%CI:0.842.84 S peryear,S only hospitals performing14/year (aOR )hospitals with ≥30 TIPS/year. Inpatient mortality :1.253.17;p<0.01) 1019/yearand (aOR:1.9, entofadmission, mode the unadjusted . Hospitals adjusted sehold he indexhe ission, TIPS/year; TIPS/year; th th d three d teaching ortality , , r r r of of r s with s 18/443 ; : 1.9, : 9

A multivariateA analyzemodel to with electpatients TIPS volumeTIPS and modeadmission (of mortalityhighest rate the for quintile (Fig(7.5%) significantlyhigher than those per ye≥20 for TIPS mortality(Appendix AdjustedTable4). odds mor for andcoagulopathy renal failurecomorbidities as we emergent admissions emergent foundincrease age,invariceal thanthosehigher per year≥20 for 2.5,(aORTIPS p during theindexduring coagulopathyandadmission in were mortality(Appendix AdjustedTable3). odds mor for

7.9%) was 7.9%) 2.7times the mortalityhigh rate the for 2). Foremergent2). admissions, unadjusted the mortali tier 4*mode, p=0.181).4*mode, tier

Author Manuscript

This article isprotected by copyright. All rights reserved. Tier 1*mode, p=0.379; Tier 2*mode, p=0.774; Tier 3* p=0.774; TierTier 1*mode, 2*mode, p=0.379; ure 3). Noure3). interaction significant was notedbetwee est of est quintile volumeTIPS 3.1)(≥30 TIPS/year; (F Hepatology Hepatology ar (aOR p=0.01).ar 1.5, ive varicealadmissionsthat bleedingfound inf and re independently re with associated oddshigher ofinp talitywithhospitalsin per119year wasTIPS als talitywithhospitalsin per119year wasTIPS sig =0.04). Similarly=0.04). analysismultivariate ofpatients ty rate thelowestfor rate tyquintile (14.2%)was 1.9 tim bleeding andbleeding infection index during the admission, dependentlyassociated with oddshigher ofinpatien mode, p=0.498; p=0.498; mode, n annualn with igure ection o nificantly es es the atient and Page 10of28 t t 10 Page 11of28 both emergentand both elective Hospitals procedures. wi Specifically,20 represented TIPS/year an annual th washospitals (6.1%),largecompared tosimilar res admission. absoluteThe adjustedindifference mort national In this ofstudy procedures,TIPShigherv DISCUSSION The selectivereferralThe hypothesis appearsexplai to ≥20 were≥20 TIPS/year exclusively teachinmetropolitan percutaneous cardiovascularpercutaneous procedures), complicati makesperfecthypothesis positsphysiciansthat and hypothesesmainTwo seekto explain volumeoutc the independentofother factors. inpatient mortalitywas patientswith for higher va buthospitalsaccounted onlyoffor 38% prall TIPS procedurespecificcausesvolumeoutcome the for re skills resulting skills betterinoutcomes. selective The quintiles performed quintiles liver transplantation. Therefor better outcomes attractbetter patients.more Inaddition, at highvolume hospitalsat liverendstagehad diseas analysis, therefore thelower mortalityinpatientw Patientsmanagement.undergoing liver transplantati more experiencedmore operators, improvedand periand patients, inclusionfor in our model.multivariate significanceofAPRDRGof risk themortality, most performedtorescuepatientswith pooroutcomes aft

Author Manuscript

This article isprotected by copyright. All rights reserved. Of thesignificant Of riskrevealed by factors the mul referral hypothesis statesreferral and/orhospitalsthat ph riceal bleeding, nosocomial bleeding, riceal andinfections, coagulo as likely as combinationrelatedtoa better of patien e, itplausibleis patientse, that benefitting b from studiesin the surgical haveliterature identified Hepatology Hepatology olume hospitalsolume deathsduringhad fewer index the n someoftheinn Hospitalsthis findings study. per ocedures. Finally,ocedures. additioninannual volumto TIPS discriminative predictiveand riskscore mortality ults for open surgical ults open for procedures (0.2%12.5%).(1) e requiringe referraltothese liver transplantation reshold beyondreshold which inpatient mortalitywas better alitybetweenrates lowvery veryvolume and high er alsoTIPS. isThis supportedlackby of the stat hospitalpersonnel, seewho patients, more develop on duringon theindex wereadmission excluded from th ≥20 were≥20 TIPS/year th exclusively teac metropolitan g hospitals and g 80%more ofhospitals than in the t on rateson and failure torescuecomplications.( from lationship. includeThese annualoperator volume (e postproceduralcare liver rather than transplantat omerelationship procedures.(26) for practiceThe etteroutcomes tivariate several t selection, ysicianswith centers for forming forming in cirrhoticin pathy, istical e, for ion 24) op two better better hing .g. in.g. 11

fluoroscopic fluoroscopic techniques for creationTIPS relyon b practicemakesperfectThe hypothesis alsocan pote inoutcomes thesepatients, distinctannual from TI as endoscopist as skill and effectiveballoon tamponad procedural andcare, recognizingor c rescuing from hepatologists,surgeons, interventionalradiologist inpatient mortalitywas primarilydriven dec by the knowledge andknowledge operator smallexperience.A study re procedural procedural intraproceduralandtime complications, creation for directvisualizationTIPS that found o experienceinexperiencedand physicians.(27) Simila tract infection) tract asindependent an risk factor for time, were time, notavailable theNRD thereforewheth in liver dysfunction.liver However, thepresenceof nosocom (<20 placed).(28)TIPS (<20 These supportfindings may t Based on ouronBased it appears results, selectiverefe the unknown. isafter TIPS either investigating either interplay the between these fa admissions (whichadmissions typically ascites)occur for acro admissionsEmergent typically(which varioccur for inpatientfactors for mortalityain sub persisted model, variceal model, bleeding coagulopathyand areinher The volumeoutcomeThe relationshipwasthat observed i improvementwill leadto likelythein care the for technical success.technical However,specific details for TI

Author Manuscript

This article isprotected by copyright. All rights reserved. analysisatelective looking emergentand admission mortalityrepresentmay modifiablea risk factor fo sepatients. f the TIPS needlethe f duringTIPS transhepaticpunctures re ision making ofisionmaking multidisciplinary (likelyincteams s, and intensivists)s, patient regarding selection,p ctors orstandardizing ctors and disseminating bestcare rral hypothesis bestrral explains our results. appea It PSprocedures,such as identity and operator proced Hepatology Hepatology ss hospital ss quintilesbut lowerinhospitals volume PSvolume. lind transhepatic portallind transhepatic vein puncturesby guided ceal bleeding)cealwere common more elective than however effectwas this inexperiencedlimited to o e, likely e, toimprovedcontribute outcomes.Future w omplicationsof care. wellThese as factors, as fac er more operatorermore experience betterinresultsoutco enttothis population toportalduehypertension a hepracticemakesperfect hypothesis as it relates rly, a studya rly,of intravascularevaluatingtheuse u ntially applythe TIPS procedure. to Traditional ialinfections(e.g. pneumonia, clostridiumdiffici vealed significant differencesvealed in significant procedure b times n this study withstudy20 this a n thresholdandTIPS/year risk r improving r luding eri rs rs that quintiles s. s. duced le, urinary le, anatomic anatomic practices ltrasound tors suchtors perators nd ork ork ure ure mes to etween Page 12of28

12 Page 13of28 less clear.First,less most patientswith emergentadmi benefit However,the higheroftransfers volumeto outcomes electivefor EVAR.(13,14) volumehascenters the toimprovepotential patient highera hadpercentage emergent admissions. of Reg survival treatmenthighvolumebenefit forata hos analysis of allpatients analysis esophagealwith of variceal b whomendoscopicandmedical management represent fi indication ascites,of TIPS (e.g. variceal bleeding mortality (35%) uncoveredfor insalvageTIPS the s coveredTIPS, it bemay unsafe to transfer the majo data is notis data available accuratelyobtainableusi nor clinical course. Finally,clinical instudyoutcomescourse. this we comparingtheaccuracydefinitions of these with me (Appendixadmissions 4),and Tablethese3 patients variceal of bleedingan as indication plac TIPS for in bleeding) selected patientsafter successfulend guidelineshave recommendedofthe use earlycovere in TIPS patientse.g.TIPSin MELD Althoughscore. thecst findingstocodingerror and reducedspecificity co bleeding, infection, bleeding, and ICUwere stays derived fro assessments)laboratory affectmay the model.Secon demonstrate discriminativegood ability,stilis it Severalareinherentlimitations toour desigstudy asadmission long as can safetransfer be arranged. lowvolume at hospitals whoderivemay benefit from

Author Manuscript

This article isprotected by copyright. All rights reserved. l possibleunmeasuredlthat confounders (e.g.clinic ,BuddChiarisydrome), totalprocedure post time, re limited re toinpatient mortalityratherthan post leeding inleeding betweentheNIS 19982005 didnotshow a ng ICD9codesng covered e.g. uncoveredvs. stent use ement in patientsinwithement bothelective emergentand n. First, theusen. administrative ofan sub database mpared toclinicalmpared assessmentswith discriminative Hepatology Hepatology oscopic therapy.(31,32)oscopic highadjustedGiven the odd pital.(19) Third, inpatientpital.(19) Third, andmortality (75%) 6w ssions likelyssions representpatients with variceal blee m priorstudies in m cohorts,nosimilar validation s TIPS centers patientsTIPS for emergentrequiringadmiss outcomes, similar to improvedhowprocesshas this etting is highettingis despiteand lack ofthe robust data rity ofrityacutely these patients.(30)sick However, r atistic thepredictivefor models developed thisin dicallevelrecord dataexist.specific patiThird,

a regionalizationregardlessofstrategy, modeof represent subseta patients of currently undergoi d, although definitionsd, conditions suchfor as vari d TIPS placementTIPS d hours(<72afterindex variceal ionalizationelectiveprocedures of higher TIPS to rstlinetreatment.(29) Second,volumeoutcomea discharge al or or al ent level ent tudies eek eek ding, for ding, for TIPS jects our jects for for analysis ecent properties s ratio s ng TIPS ng TIPS ceal , ions isions

In conclusion, hospitalsIn performing 20more than TI However,mortality.similar studieson proced other outcomes is used.(1,33)is outcomes andinpatient 30daymortality differenceno andin regionalizationpatients elerequiringstrategy for withhospitals lower volume.annualTIPS Standardiz volumehospitals.

Author Manuscript

This article isprotected by copyright. All rights reserved. ctiveleadmayimprovementTIPS to ofoutcomes l at Hepatology Hepatology thevolumeoutcome relationshipwhenof either the ures have previouslyures correlationbetweenshown good PSperyear have inpatientlower mortalitycompared ing patientselection, ing periproceduralcare, and a ow se se Page 14of28 to

14 Page 15of28 volumenonelective for admissions. 3:ObservedFigure mortalityand indication rate fo volumeelective admissions.for Figure 2:ObservedFigure mortalityand indication rate fo measures congestivefor heart failure, coagulopathy volume.hospital MortalitywasTIPS rate adjusted f 1:AdjustedFigure inpatient patmortalityrate for LEGENDS: FIGURE admission, variceal admission, bleeding, infection,diabetes,

Author Manuscript

This article isprotected by copyright. All rights reserved. ientsundergoing inTIPS according2013 to quintile quintiles ofannual quintiles volume,procedure comorbidAHRQ Hepatology Hepatology r placementbyr TIPS annualhospital quintiles of TI r placementbyr TIPS annualhospital quintiles of TI or:age,presenceof alcoholic disease,liver emerg and renal failure, andrenal and failure, income perpatientquartile ofannual PS PS zip code. ent ity 15

6 DariushniaHaskalZJ,MidiaSR, M,Martin LG, 16. 14. Ilonzo N, Egorova IlonzoNowygrod NN, N, R. Interhospita 14. 12. Brooke BS, Perler Brooke BA, Makary Dominici F, MA, P 12. 11. Birkmeyer JD, BirkmeyerFinlayson BirkmeyerEV, CM. Volu 11. 3 HillMcPheeMessina JS, JT, LM, EslCioccaRG, 13. 15. Perry BC, KwanBC, Perry PortosystemicSW. StablShunts: 15. 1. Birkmeyer JD, BirkmeyerSiewers Finlayson AE, Stukel EVA, 1. REFERENCES . GonzalezZM,Abdelsattar AA, JB,DimickDev S, 2. 9. Dua A, Furlough DuaA, CL, Ray SharmaS,Upchurch H, G 9. 10. Dua A, Romanelli DuaA, M, Upchurch Pan GR, HoodJ, D 10. 8. Holt PJE, Poloniecki PJE, HoltJD, D,GerrardLoftus IM, 8. 4. Singh JA,Ramachandran Singh Does R. hospitalvolume 4. . ShuhaiberAJ,IsaacsJ, Sedrakyan EffectA. The 3. 7. Arora S,Panaich Arora PatelSS, N,N, PatelLahewala 7. 6. Badheka AO, Patel BadhekaAO, NJ,P, Singh V, Grover Patel 6. 5. Badheka AO, Patel BadhekaAO, NJ, PatelPanaichSS, SV, Jha 5. After the After TransjugularPortosystemic Intrahepatic S benefits of benefits theLeapfroginitiative. 2001;Surgery. California hospitals California Leapfrog meeting evidencebase repair: evidencearepair: shifthighvolumeof centers to Vasc. Surg. Vasc. 2008;47:1155–1156; discussion 1163–1164 carotid intervention. carotid Vasc.J.Surg. 2016; Guidelines TransjugularPortosystefor Intrahepatic Vasc. Surg. Vasc. 2016;63:859–865.e2. Surgical Mortality Surgical thein States.UnitedN.J Engl. mortalityafterabdominalrates aorticaneurysm rep 2007;94:395–403. the relationshipbetween volumeoutcomeand in abdo 1346. 1346. Mitraland ValveProcedures: Population ASurgical HighriskMortality Ann.After 2015;Surg.Surgery. [2006 to 2011]). Am. [2006 to Cardiol.J. 2015;116:791–800. ofOutcomesLower Extremity Endovascular Interventi institutional volumeinstitutionalpercutaneous on coronary inter Outcomes ofOutcomes TranscatheterAorticValve Implantation arthroplastyshoulder theinUS?Arthritis Res Care (20052009).Circulation. 2014;130:1392–1406.

Author Manuscript

This article isprotected by copyright. All rights reserved. Hepatology Hepatology Thompson MM. Thompson Metaanalysissystematicand review of Walker KalvaTG, et al. SP, Walker QualityImprovement lintact transfer abdominalfor aorticaneurysm rep 130:415–422. . Med. 2002;346:1128–1137.. of CenterVolume of on InHospital AoMortality After S, Solanki S,S, et al. ImpactofHospital onVolume ami MH. ami Regionalizationabdominal of aorticaneurys ronovost PJ.Reductionronovost inhospital of mortalityamo N, Arora N, S,et al. Impactofannualoperator and in the endovascular the in Vasc.Surg.era. J. 2008;48:29– me standards me highriskforprocedures: surgical pot Birkmeyer JD,Birkmeyer AA.Ghaferi Timetoreadmission and e Utilizatione ImprovedOutcomes,and DecadesTwo TA, Lucas FL,Batista al.TA, I, HospitalVolume et an 262:53–59. mnani S,Singhmnani V, etal. EffectofHospital Volume . 2015;67:885–890. . mic Shunts. mic Vasc.J. Radiol.Interv.2016;27:1–7. , HodgsonKJ, Predictors, et al. ofpooroutcome aft air. J. Vasc.J. air. 2014;60:1446–1451.Surg. hunt. Coll.Radiol.Am.J. hunt. 2015;12:1427–1433.JACR. ventionoutcomes:a 5year experienceStatesUnited predictoutcomes and aftercomplications total Based Study.Based Ann. 2015;100:1340–Surg.Thorac.

R,DesaiSS. effectofhospitalThe on factors d standardsdabdominal aorticaneurysmfor J repair. . Am. Cardiol.J.. 2015;116:587–594. ons (Insights ons from theNationwide SampleInpatient minal aorticaneurysmminal surgery. Surg.J.Br. . . air.J. Page 16of28 rtic rtic d ng ential on 36. 36. er m m

. .

Page 17of28 4. Ghaferi AA, Ghaferi Birkmeyer JD,DimickHospital JB. 4. 5. Ghaferi AA, Ghaferi Birkmeyer JD,DimickVariation JB. 5. 3. Vergara M, ClèriesM, Vergara Vela Bustins M,E, Mique 3. 6. Luft HS, HuntHS,SS, Luft Maerkivolumeoutcom SC. The 6. 9 MyersPapayRP,Shaheen KD, AAM, KaplanReGG. 9. 2 MyersQuanH,RP,Hubbard JN,Shaheen AAM, Kap 2. 1. Healthcare Costand Healthcare Utilization ElixhProject. 1. Costand Healthcare Utilization N Project. The 0. 8 MellingerJL,Richardson MathurAK,CR, Volk M 8. 7. Trivedi PS, Rochon TrivediPS, PJ, Ryu Durham JD, RK. Nati 7. 9 GarciaTsaoSanyal G, ND,AJ, Grace Carey P W, 9. 8. Pillai AK, Andring B,Andring PillaiFaulconer AK, N,Reis SP, Xi 8. 7 MarquardtS,Rodt H,Rosenthal T, F,Me Wacker 7. 1. Tripathi D, TripathiStanley AJ,Hayes Patch D,PC, Mil 1. 0. Vangeli M,Burroughs Vangeli Patch D, AK. Salvagetips 0. 2 GarciaTsao Abraldes G, BerzigottiJ,A, Bosch 2. Association for the StudyAssociation the for ofLiverPracti Diseases, specific complicationsspecific of Livercirrhosis. Int.Off Care. 2011;49:1076–1081. Care. Engl. J. Med.J.Engl. 2009;361:1368–1375. with cirrhosis:with differresults across riskadjustme [cited 2016 Jun[cited 30];Available https://www.hcu from: of Esophageal Varicealof theinBleedingUnited State 2015;13:577–584; e30.quiz Clin.mortalitycirrhosis. for Gastroenterol.Hepat PortosystemicCardiovasc.Shunt. Intervent.Radiol. onFactorsSkill theSuccess Duration and of Fluoro referral patterns?Health referral Res.Serv.1987;22:157–18 2016 Jun 30];AvailableJun 2016 https://www.hcupus.ahfrom: 2002;37:703–704. http://www.jvir.org/article/S105104431501249X/abstr 2016Radiol. [Internet]. Apr2016[cited25];0. Ava Portosystemic Intrahepatic Creation Shunt Using the cirrhosis. Hepatol.cirrhosis. Baltim.Md. 2007;46:922–938. Gastroenterology.Prevention and managementof gast with Conventionalwith 109in Technique Patients. VasJ. duringVein Access IntrahepaticTransjugular Portos Study of Liver Diseases.Studyof Baltim.Md.Hepatol. 2016; Stratification,Diagnosis Managementand 2016 Pra ofvariceal management inhaemorrhage cirrhotic pat

Author Manuscript

This article isprotected by copyright. All rights reserved. Hepatology Hepatology auser Comorbidity Software,auser 3.7Version[Internet]. Y,Iyamuetal. I, Utility of IntravascularUSGui ationwideReadmissionsDatabase Overview [Internet] lson C, Mehrzad C, et al. UKlson H, guidelineson the . J. Int..J.Assoc.Study Liver. 2013;33:828–833. J.Hypertensive Portal Cirrhosis:inBleeding Risk volumeand failure to rescuewithsurgery highrisk in hospital mortalityassociated with inpatient su l M,R.lHospital Campo mortalityoverpatieintime ntmethods.Hepatol. Baltim. Md. 2009;49:568–577. for uncontrolled variceal for bleeding. Hepatol.J. ol. Off. Clin.ol. Off. Pract.Am.J. Gastroenterol.Assoc. L. Variation L. amongUnited hospitalsStates inpatin yer BC. ImpactofAnatomical,yerBC. Procedural, and Opera ilable from: onalTrendsand ofOutcomes Transjugular ce ParametersCommittee ce oftheAmerican College of e practicemakesperfectrelationship: orselective ractice Guidelinesractice Committeeof theAmerican pus.ahrq.gov/nrdoverview.jsp scopyGuided TransjugularIntrahepatic 2015;38:903–912.

s. Clin.s. Gastroenterol.Hepatol. 2008;6:789–798. c. Radiol.Interv.c. JVIR.2016; ShuntystemicCreation: Retrospective Comparison 2. 2. Nationwide Vasc.J. Sample. Inpatient Interv. rq.gov/toolssoftware/comorbidity/comorbidity.jsp act ients.2015;64:1680–1704. Gut. ctice byGuidance theAmerican Association the for lan GG. Predicting inhospitalPredictinglanGG. mortalitypatientin lationshipBetween HospitalVolume and Outcomes roesophagealvaricesvaricealand hemorrhage in dedPortal rgery.N. [cited

.Med. nts with nts ient . s s tor 17

3. Rosenthal GE, BakerGE, RosenthalDW,Norris LE,DG, Har Way 3.

2000;34:1449–1468. 30daystandardizedand hospital implicamortality:

Author Manuscript

This article isprotected by copyright. All rights reserved. Hepatology Hepatology tions for profiling tions hospitals.for Serv.Health Res. per DL,Snow per Relationships RJ. between inhospital Page 18of28 18 Page 19of28 Age (year) Age APR CMAHRQ CMAHRQ CMAHRQ Renal Acute Failure Ascitesorhydrothorax useAlcohol Hospitalteaching status Hospitalownership Hospitalbedsize hospitals of N 1280 167 332 2225 228 19 213 19.1 799 2885 211 82.0 20.2 imbalanceElectrolyte 904 243 79.7 Infection 1889 18.0 958 994 74.7 HCC HRS 4128 HE SBP Bleeding Weekendadmission admission Emergent stayICU Medianoflengthstay (days) Diabetes abuseSubstance (nonalcohol) 125 byIncomequartile 13.2 146 12.9 155 10.5 Insurancestatus 583 ofliver Etiology disease Female Inpatient mortality admissions of N DRGRisk ofMortality 71. 5±13 71. 5±11 61. 5±. 0.0 57±9.9 56±10.7 57±11.1 57±11.7 58±11.3 57±10.9 CF 811 2593 308 CRF Coag. CHF 16 95 5 4. 43 38 4 4. 21 96 418 29.6 291 41.3 444 43.8 483 45.8 550 39.5 2186 oeaelklho fdig1355 Moderateofdying likelihood xrm ieiodo yn 1801 likelihood Extreme ofdying Metropolitannonteaching ao ieiodo yn 2159 Majorofdying likelihood 200 Minorofdying likelihood Metropolitanteaching 1466 Private proprietaryPrivate ocasseiid1 02 00 02 00 1 . 0 1.1 11 0.0 0 0.2 2 0.0 0 0.2 13 specified Noclass Nonmetropolitan 3803 $64,000or more Private nonprofitPrivate $48,00063,999 $38,00047,999 Government Nocharge Author Manuscript$137,999 Medicare Alcoholic Medicaid

Selfpay Medium NAFLD Private

L a r g e Small Other Viral 75 14 6 3. 30 18 4 3. 29 54 2 3 422 25.4 1319 249 32.4 349 31.8 350 30.4 365 31.4 1735 3 448 43.1 423 33.5 360 36.0 397 36.9 443 37.5 2071 17 06 7 2. 27 34 0 1. 27 11 9 1 196 21.1 207 18.7 200 23.4 257 23.2 277 20.6 1137 09 90 9 1. 27 88 0 1. 11 64 7 2 278 16.4 161 18.9 204 18.8 207 16.5 199 19.0 1049 45 61 8 2. 38 89 6 2. 37 13 7 2 2 271 324 31.3 26.5 307 260 24.4 32.8 263 353 28.9 28.5 318 314 23.8 28.7 286 345 26.1 28.9 1445 1596 33 43 7 2. 20 36 6 2. 24 48 0 2 301 24.8 244 24.1 260 23.6 260 23.1 278 24.3 1343 48 52 2 5. 53 84 2 4. 32 69 5 3 458 36.9 362 48.6 523 48.4 533 51.7 622 45.2 2498 42 61 9 2. 25 14 4 2. 36 32 4 2 347 33.2 326 22.5 241 21.4 235 24.5 293 26.1 1442 56 83 9 2. 35 04 2 2. 22 26 9 3 391 22.6 222 29.9 322 30.4 335 24.6 296 28.3 1566 12 87 3 3. 48 89 2 3. 34 50 9 4 497 35.0 344 39.7 426 38.9 428 36.6 437 38.7 2132 426 7.7 105 8.8 106 9.6 101 9.4 28 2.8 86 7.4 7.4 86 2.8 28 9.4 101 9.6 106 8.8 105 7.7 426 379 6.9 83 6.9 73 6.6 104 9.7 78 7.9 41 3.5 3.5 41 7.9 78 9.7 104 6.6 73 6.9 83 6.9 379 8 65.0% 288 2 1. 25 29 3 1. 18 28 9 1. 11 15 181 19.7 193 12.8 138 12.3 135 22.9 275 16.7 922 4 1. 14 . 12 . 9 84 1 2. 19 11.1 129 21.9 215 8.4 90 9.3 102 8.6 104 11.6 640 1 70.2% 311 2 27.8% 123 2 50.1% 222 348 6.3 99 8.2 77 7.0 54 5.0 44 4.5 74 6.3 6.3 74 4.5 44 5.0 54 7.0 77 8.2 99 6.3 348 0 46.5% 206 9 13.3% 59 5 .% 4 .% 11 0 .% 00 0 0.0% 0 0.0% 0 0.0% 0 1.1% 1 5.0% 14 3.4% 15 2 .% 9 04 2 .% 26 0 .% 0.0% 0 0.0% 0 2.6% 1 2.1% 2 10.4% 29 7.2% 32 3 16.5% 73 Full Full cohort n This article isprotected by copyright. All rights reserved. 7 (313) 8 (413) 8 (414) 7 (312) 7 (313) 7 (31 7 (313) 7 (312) 7 (414) 8 (413) 8 (313)7 n=5529 443 39 1 2. 24 67 5 2. 19 02 6 2. < 22.3 260 20.2 199 23.5 253 26.7 294 26.0 313 23.9 45 7 2. 23 02 7 2. 27 51 3 28.3 330 25.1 247 25.8 277 20.2 223 23.1 278 24.5 65 0 2. 21 64 7 2. 28 42 6 3. < 31.4 367 24.2 238 25.1 270 26.4 291 25.0 300 26.5 47 5 1. 17 42 5 1. 11 23 2 1. < < 18.9 42.0 221 490 12.3 45.0 121 442 14.6 47.9 157 515 14.2 51.0 157 562 12.9 48.6 155 584 14.7 46.9 88 7 6. 65 30 2 6. 65 87 2 7. < 79.5 928 68.7 675 67.6 727 63.0 695 64.7 778 68.8 02 1 4. 47 23 8 4. 33 19 4 3. < 38.0 0 444 22.0 31.9 257 313 26.0 45.5 255 489 22.0 42.3 237 < 467 39.9 23.0 42.6 466 253 512 45.3 23.1 < 40.2 445 278 23.7 55.7 23.2 277 599 38.0 60.6 373 668 35.2 58.8 379 707 41.2 52.2 454 33.8 406 34.2 91 9 4. 40 72 7 3. 43 10 7 40.7 475 41.0 403 35.2 378 37.2 410 41.0 493 39.1 26 7 3. 47 87 8 3. 31 06 1 26.8 313 30.6 301 35.5 382 38.7 427 31.5 378 32.6 . 7 61 4 . 6 56 0 . 6 52 0.10 5.2 61 4.1 40 5.6 60 6.7 74 6.1 73 5.6 . 2 19 2 . 3 31 5 . 4 38 <0.01 0.02 3.8 6.1 0.07 44 71 3.0 4.6 7.8 35 45 77 5.5 3.1 6.6 54 33 71 4.4 2.0 4.8 47 22 53 4.1 1.9 5.0 45 23 60 3.9 3.0 6.0 47 4.1 3.6 52 4.3 40 3.6 38 3.5 21 2.1 49 4.2 4.2 49 2.1 21 3.5 38 3.6 40 4.3 52 3.6 % ComplicationsofInpatient Hospitalizations

ComorbiditiesandRisk ofMortality n=1202(22%) 1 3.% 1 53.7% 45.3% 51 43 39.6% 55.4% 110 70.5%154 67 69.1% 77.9%192 74 54.7% 152 HospitalizationCharacteristics 6 01 1 13.7% 13 15.8% 20.1% 15 56 10.8% 20.0%30 19 34.9% 97 n CharacteristicsHospitals of ComplicationsofCirrhosis Patient DemographicsPatient 278 14 Hepatology Hepatology % n=1102(20%) n 59 95 %

n=1076(19%) 0 76.9% 30 74.4% 29 87.2% 34 n 77 1 .% 0.0% 0 5.6% 23.1% 1 9 7.7% 3 17.9% 7 10.3% 4 1019 39 43 8 6. 74 72 <0.01 67.2 784 69.6 683 74.3 <0.01 6.0 70 8.9 88 11.6 % 8 6 1. 13 40 <0.01 14.0 163 16.7 164 .8

n=982(18%) 8 100.0% 18 61.1% 11 97.7% 17 n 00 0 0.0% 0 0.0% 0 15.4% 33.3% 6 2 5.6% 1 2029 18 %

n=1167(21%) 3 100.0% 13 92.3% 12 84.6% 11 n 7.7% 1 0.0 0.0 ≥30 13 35.8 2) % 6.2 6.2 <0.01 8.4 6.8 6.8 3.8 3.8 3.2 3.2 7.8 5.8 5.8 9.3 9.3 9.7 9.7 3.5 3.5 2.6 2.6 .5 .5

p valuep <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 0.03 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 .20 .20

Refined Patient Diagnosis Related Groups. AHRQcarcinoma, CM:Healthcare Agency for Research peritonitis,bacterial HE: encephalopathy, hepatic 2:Multivariate Table modelinpatient mortality for 1: Table Inpatient characteristicsmortality,of pa Livertransplanthospital Hospitalurbanrural location procedure volume.procedureNAFLD: Nonalcoholic fattyliver ag erpltn> ilo eiet 6 59.4% 263 Largemetropolitan >1 millionresidents ml erpltn< ilo eiet 6 37.2% 165 Small <1 metropolitan residents million Other riskOther factors quartileIncome for patient code zip AHRQcomorbidity measures Annualprocedure volume AllPatients

Age (peryear Age 1 increase) Congestiveheart failure Alcoholicliverdisease irpltn1 34 1 50 1 .% 00 0 .% 0 0.0% 0 0.0% 0 1.1% 1 5.0% 14 3.4% 15 Micropolitan

Author Manuscriptadmission Emergent

Variceal $64,000or more 20 10

$48,000 $38,000 ≥30 Coagulopathy 5 1 29 19 Renalfailure 9 4

aOR 0.8 3.9 2.3 1.4 1.0 1.0 0.6 0.5 0.9 1.5 2.0 1.4 1.4 1.9 2.0 1.9

0 3.% 5 36.8% 62.1% 35 59 36.3% 58.6% 101 163

04 5 .% 5 38.5% 15 5.3% 5 0.4% 1 tients,hospitalizations and hospitals eachfor qui after TIPS hospitalization.after TIPS 0.60 1.06 1.46 0.85 0.84 1.17 1.25 1.21 0.55 3.00 1.65 0.88 0.78 1.01 0.38 0.36 Reference Reference HRS: hepatorenalHRS: syndrome, hepatocellularHCC: Hepatology Hepatology 95%CI disease, ICU: Intensivecareunit,SBP: Spontaneous

1.29 2.21 2.62 2.31 2.34 3.00 3.17 3.01 1.04 5.18 3.21 1.04 1.40 1.04 0.90 0.82

and QualityComorbidity Measures,APRDRG: All

p value p 0 01<0. 01<0. <0.0 <0.0 01<0. 01<0. 0.51 0.03 0.18 0.19 0.01 0.01 0.08 0.16 0.76 0.01

1 1

9 48.7% 51.3%19 20 c - statistic 0.776

5 83.3% 15 50.0% 50.0%9 9

3 100.0% 13 92.3% 12 7.7% 1 ntileannual of 0.0%

Page 20of28 <0.01 0.07

Page 21of28

Author Manuscript

Page 22of28 Page 23of28

Author Manuscript

303.90-93 OTHER AND UNSPECIFIED ALCOHOL DEPENDENCE DEPENDENCE ALCOHOL UNSPECIFIED AND OTHER MENTAL DISORDERS UNSPECIFIED ALCOHOL-INDUCED INTOXICATION ALCOHOL ACUTE DISORDERS ALCOHOL-INDUCED OTHER 303.90-93 SLEEPDISORDERS ALCOHOL-INDUCED 303.00-03 WITHDRAWAL ALCOHOL DISORDERS MENTAL ALCOHOL-INDUCED SPECIFIED OTHER 291.9 DELUSIONS DISORDERWITH PSYCHOTIC ALCOHOL-INDUCED 291.89 ALCOHOLINTOXICATION IDIOSYNCRATIC 291.82 HALLUCINATIONS DISORDERWITH PSYCHOTIC ALCOHOL-INDUCED 291.81 DEMENTIA PERSISTING ALCOHOL-INDUCED 291.8 DISORDER AMNESTIC PERSISTING ALCOHOL-INDUCED 291.5 DELIRIUM WITHDRAWAL ALCOHOL 291.4 291.3 LIVER OF DISORDER UNSPECIFIED 291.2 HEPATICINFARCTION 291.1 OF LIVER CONGESTION PASSIVE CHRONIC 291.0 Alcohol use CIRRHOSIS BILIARY PRIMARY 573.9 573.4 DEFICIENCY ALPHA1ANTITRYPSIN 573.0 LiverDisease Chronic Other OFCOPPERMETABOLISM DISORDERS 571.6 Non-Specific Cirrhosis 273.4 HEPATITIS AUTOIMMUNE Deficiency Alpha-1-Antitrypsin HEPATITIS VIRUS INCLUSION CYTOMEGALIC 275.1 OF ALCOHOL W/OMENTION LIVERDISEASE UNSPECCHRONIC Wilson's Disease 571.42 HEPATITIS CHRONIC OTHER Autoimmune Hepatitis HEPATITIS CHRONIC UNSPECIFIED 078.5 571.9 571.49 COMA MENTION HEP WITHOUT VIRALHEPATITIS SPECIFIED OTHER 571.40 COMA CWITHHEPATIC HEPATITIS UNSPECIFIEDVIRAL Hepatitis Non-Specific orOtherSpecified COMA CW/OHEPATIC HEPATITIS UNSPECIFIEDVIRAL Diseases Liver Other COMA MENTION HEPATIC HEPATITIS CWITHOUT CHRONIC 070.59 COMA HEPATIC MENTION CWITHOUT HEPATITIS ACUTE 070.71 COMA HEPATIC HEPATITIS CWITH CHRONIC 070.70 COMA HEPATIC CWITH HEPATITIS ACUTE 070.54 070.51 W/HEP DELTA COMACHRN HEP HEP BW/OMENTION VIRAL 070.44 DELTA W/OHEP CHRN HEPBW/OCOMA VIRAL 070.41 HEP DELTA W/O ACUT/UNS HEPBW/OCOMA VIRAL Hepatitis C HEPATITISDELTA W COMA, WHEPATIC HEPATITIS B,CHRONIC, 070.33 DELTA HEP W/OMENTION HEPBW/HEPCOMACHRN VIRAL 070.32 W/OHEPDELTA HEPBW/HEPCOMAACUTE/UNSPECIFIED VIRAL 070.30 070.23 DISEASE LIVER NON-ALCOHOLIC CHRONIC OTHER 070.22 ALCOHOL OF MENTION LIVER WITHOUT OF CIRRHOSIS 070.20 Hepatitis B DAMAGE LIVER UNSPECIFIEDALCOHOLIC Viral Disease OFLIVER CIRRHOSIS ALCOHOLIC 571.8 HEPATITIS ACUTEALCOHOLIC 571.5 LIVER FATTY ALCOHOLIC Liver Disease Fatty Non-Alcoholic 571.3 571.2 571.1 LIVER TRANSPLANTED OF COMPLICATIONS 571.0 OFLIVER TRANSPLANT OTHER Liver Disease Alcoholic TRANSPLANT LIVER AUXILIARY Disease Liver of Etiology 996.82 Diagnosis Description 50.59 50.51 Procedure Transplant Liver Code TABLES APPENDIX

Author Manuscript

Page 24of28 Page 25of28 51 ESOPHAGOGASTRODUODENOSCOPYBIOPSY WITH CLOSED [EGD] INTESTINE OF SMALL ENDOSCOPY OTHER BLEEDING OR DUODENAL GASTRIC OF CONTROL ENDOSCOPIC OF TISSUE OR LESION OF DESTRUCTION OR EXCISION ENDOSCOPIC OF TISSUE OR LESION OF DESTRUCTION OR EXCISION ENDOSCOPIC 45.16 45.13 UNSPECIFIED TRACT, GASTROINTESTINAL OF HEMORRHAGE 44.43 BLOODINSTOOL 43.41 HEMATEMESIS 42.33 ANUS AND OF RECTUM HEMORRHAGE (endoscopy) Procedure 578.9 HEMORRHAGE ESOPHAGEAL 578.1 VARICES IND ESOPHAGEAL BLEEDING VARICES WITH ESOPHAGEAL 578.0 569.3 530.82 456.20 456.0 COMPLICATION DIABETES WITHUNSPECIFIED Diagnosis DIABETES WITHOTHER Variceal Bleeding DISORDERS CIRCULATORY DIABETES WITHPERIPHERAL 250.90-93 MANIFESTATIONS DIABETES WITHNEUROLOGICAL 250.80-83 DIABETES WITHOPHTH 250.70-73 DIABETES WITHRENALMANIFESTATIONS 250.60-63 DIABETES WITHHYPEROSMOLARITY 250.50-53 DIABETES WITHKETOACIDOSIS 250.40-43 COMPLICATION OF MENTION WITHOUT DIABETES MELLITUS 250.20-23 USE SUBSTANCE COUNSELING, 250.10-13 COCAINE BY POISONING (HALLUCINOGENS) PSYCHODYSLEPTICS BY POISONING 250.00-03 Diabetes NARCOTICS ANDRELATED OPIATES OTHER BY POISONING V654.2 METHADONE BY POISONING 970.81 HEROIN BY POISONING 969.6 UNSPECIFIED OPIUM (ALKALOIDS), BY POISONING 965.09 965.02 O CHILDBIRTH, PREGNANCY, COMPLICATING DEPENDENCE DRUG 965.01 ABUSE DRUG OR UNSPECIFIED MIXED, OTHER, 965.00 ABUSE SYMPATHOMIMETIC ACTING OR RELATED AMPHETAMINE 648.30-34 ABUSE COCAINE 305.90-93 ABUSE OPIOID 305.70-73 ABUSE OR ANXIOLYTIC SEDATIVE,HYPNOTIC 305.60-63 ABUSE HALLUCINOGEN 305.50-53 CANNABIS ABUSE 305.40-43 DEPENDENCE DRUG UNSPECIFIED 305.30-33 TYPE OPIOID EXCLUDING DEPENDENCE OFDRUG COMBINATIONS 305.20-23 OTHER DRUG ANY TYPE DRUGWITH OPIOID OF COMBINATIONS 304.90-93 DEPENDENCE DRUG SPECIFIED OTHER 304.80-83 DEPENDENCE HALLUCINOGEN 304.70-73 DEPENDENCE OTHERPSYCHOSTIMULANT AND AMPHETAMINE 304.60-63 CANNABIS DEPENDENCE 304.50-53 DEPENDENCE COCAINE 304.40-43 DEPENDENCE OR ANXIOLYTIC SEDATIVE,HYPNOTIC MENTALDISORDER DRUG-INDUCED UNSPECIFIED 304.30-33 TYPE OPIOID DEPENDENCE 304.20-23 DISORDERS MENTAL DRUG-INDUCED SPECIFIED OTHER 304.10-13 SLEEPDISORDERS INDUCED DRUG 304.00-03 MOODDISORDER DRUG-INDUCED 292.9 AMNESTIC DISORDER PERSISTING DRUG-INDUCED 292.89 DEMENTIA PERSISTING DRUG-INDUCED 292.85 DELIRIUM DRUG-INDUCED INTOXICATION DRUG PATHOLOGICAL 292.84 292.83 HALLUCINATIONS WITH DISORDER PSYCHOTIC DRUG-INDUCED 292.82 DELUSIONS WITH DISORDER PSYCHOTIC DRUG-INDUCED WITHDRAWAL DRUG 292.81 292.2 292.12 CARDIOMYOPATHY ALCOHOLIC 292.11 GASTRITIS ALCOHOLIC POLYNEUROPATHY ALCOHOLIC 292.0 alcohol) than use (other Substance 535.30-31 ABUSE ALCOHOL 425.5 357.5 305.00-03

Author Manuscript

ALMIC MANIFESTATIONS ALMIC SPECIFIED MANIFESTATIONS MANIFESTATIONS SPECIFIED ISEASES CLASSIFIED ELSEWHERE, WIT CLASSIFIEDELSEWHERE, ISEASES Hepatology Hepatology DEPENDENCE R THE PUERPERIUM R THEPUERPERIUM DRUG DRUG H BLEEDING H BLEEDING STOMACH STOMACH ESOPHAGUS

pedx al 1 it f nentoa Casfcto o Diseas of Classification International (LOCAL) (CENTRAL) HYPOTHERMIA of Modification (ICD-9-CM)codes List AQUAPHERESIS 1 Table Appendix TUBE TRACHEOSTOMY OF REMOVAL RESUSCITATION CARDIOPULMONARY TUBE REPLACEMENTTRACHEOSTOMY OF 99.81 99.78 SUBSTANCE NUTRITIONAL OF CONCENTRATED PARENTERALINFUSION VENTILATION MECHANICAL INVASIVE OTHER 99.60-63 SUBSTANCES NUTRITIONAL OF CONCENTRATED ENTERALINFUSION 97.37 97.23 TECHNIQUE OTHER BY OUTPUT OFCARDIAC MONITORING 96.70-72 TECHNI CONSUMPTION OXYGEN BY OUTPUT OFCARDIAC MONITORING RESPIRATORYTHERAPY 99.15 MONITORING WEDGE ARTERY PULMONARY 96.6 CENTRAL VENOUS PRESSUREMONITORING 93.90,91,99 MONITORING GAS BLOOD INTRA-ARTERIAL CONTINUOUS 89.68 PERITONEALDIALYSIS 89.67 HEMODIALYSIS 89.64 [ECMO] OXYGENATION EXTRACORPOREAL MEMBRANE 89.62 CATHETERIZATION ARTERIAL 89.60 PACEMAKER SYSTEM TRANSVENOUS TEMPORARY OF INSERTION 54.98 39.95 TRACHEOSTOMY TEMPORARY 39.65 AGENTOF VASOPRESSOR INFUSION SYSTEM ASSIST &CIRCULATORY OF HEART IMPLANT 38.91 37.78 37.61,62,68 31.1 00.17 UNSPECIFIED PYELONEPHRITIS, (Diagnosis) SHOCK NOTSPECIFIED SITE INFECTION, TRACT URINARY Procedure 785.50,51,59 ICU stay SEPTICEMIA 590.80 SEPTICSHOCK 599.0 tractinfection Urinary SEPSIS 038.0-9 DIFFICILE CLOSTRIDIUM TO DUE INFECTION INTESTINAL CHOLANGITIS 785.52 BACTEREMIA 995.91-92 Sepsis CELLULITIS 008.45 PNEUMONIA 576.1 HYPOPOTASSEMIA 790.7 ACIDOSIS 682.2-9 HYPONATREMIA AND/OR HYPOSMOLALITY 481-486 Infection 276.8 FAILURE ACUTERENAL PRIMARY OFLIVER, NEOPLASM MALIGNANT 276.2 SYNDROME HEPATORENAL 276.1 Electrolyte imbalance BACTERIALPERITONITIS SPONTANEOUS 584.5-9 155.0 ENCEPHALOPATHY OTHER HEPATICENCEPHALOPATHY 572.4 567.23 THORACENTESIS complications Other DRAINAGE ABDOMINAL PERCUTANEOUS 348.39 572.2 EFFUSION UNSPECIFIEDPLEURAL encephalopathy Hepatic 34.91 OVERLOAD FLUID OTHER 54.91 ASCITES OTHER drainage) (percutaneous Procedure TUBERCULOUS EXCEPT EFFUSION, FORMSOF SPECIFIED OTHER 511.9 276.69 COLONOSCOPY 789.59 511.89 Diagnosis hydrothorax and Ascites 45.23

Author Manuscript

This article isprotected by copyright. All rights reserved. Hepatology Hepatology s Nnh eiin Clinical Revision, Ninth es, QUE QUE S Page 26of28 Page 27of28 Infection Yes vs no 4.4 3.73-5.31 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 3.73-5.31 0.07 4.24-6.91 3.20-4.64 1.90-2.71 6.46-9.38 0.97-2.08 4.4 0.15 5.4 3.9 <0.01 2.3 7.8 0.95-1.39 1.4 0.28 2.47-3.65 0.19 0.50 1.1 3.0 0.63 0.89-1.39 0.67-1.88 Yes vsno <0.01 0.74-5.20 0.86-1.78 1.1 1.1 Yes vsno 0.81-1.47 Yes vsno Yes vsno Yes vsno 0.81-5.78 Appendix Table2Crude OddsRatiosforInpatientMortality Yes vsno 2.0 <0.01 Electrolyte imbalance 1.2 1.1 1.11-2.12 Acute RenalFailure Yes vsno 2.2 1.51-2.78 1.77-3.20 Infection 12.57-20.45 1.72-3.11 Hospitalizations Yes vsno of Inpatient Complications Ascites orhydrothorax <0.01 <0.01 vsnon-metropolitan teaching Metropolitan 1.5 16.0 HRS <0.01 vsnon-metropolitan non-teaching Metropolitan 2.1 2.4 HE Transpl 2.3 0.59-0.88 1.01-1.03 SBP vs privateproprietary nonprofit Private 1.45-2.15 Variceal bleeding vsprivateproprietary Government of cirrhosis Complication Liver transplantstatus vslarge Medium 0.7 vs Small large 1.0 <0.01 1.8 vs ≥30TIPS/year 20-29TIPS/year 0.81-1.37 vs ≥30TIPS/year 10-19TIPS/year 0.55-0.89 location Hospital urban-rural 0.84-1.36 vs≥30TIPS/year 5-9 TIPS/year 0.47-0.81 vs≥30TIPS/year 1-4 TIPS/year 0.49-0.83 0.48-0.76 <0.01 Hospital teachingstatus 1.1 0.7 0.63-1.31 1.1 0.6 1.12-2.03 0.64-1.01 Hospital ownership 0.6 0.6 <0.01 <0.01 0.88-1.38 Yes vsno Hospital bed-size 0.9 1.00-1.02 1.5 0.63-0.91 0.8 vsweekday Weekend 1.1 1day increase Emergentvs vs $64,000ormore $48,000-63,999 1.0 volume Annual procedure Yes vsno vs $64,000ormore $38,000-47,999 0.8 Hospital vs$64,000ormore $1-37,999 ICU stay Admission day Elective admission Other vsalcoholic Length ofstay vsalcoholic Viral Characteristics NAFLD vsalcoholic Hospitalization Diabetes vsMedicare No-charge alcohol) Self-pay vs Medicare Substance abuse(non- vsMedicare Private vsMedicare Medicaid code thezip Income rangefor year 1 increase Femalevsmale Etiologyof liverdisease Primary payer insurance Gender Age Patient

Author Manuscript <0.01 micropolitan small metropolitan <1millionresidents vs micropolitan 0.25 large metropolitan >1millionresidents vs 0.55-1.17 0.8 p 95%CI OR Yes vsno This article isprotected by copyright. All rights reserved. ant hospital vsnon-transplanth ant hospital elective 3.1 2.34-4.00 <0.01 <0.01 2.34-4.00 3.1 elective Univariate analysis analysis Univariate Hepatology Hepatology ospital 0.7 0.55-0.79 <0.01 <0.01 0.55-0.79 0.7 ospital . 07-.2 0.77-5.42 0.75-5.27 2.0 2.0

Appendix Table 4: Multivariate model forinpatient mortalityaf Appendix Table4:Multivariatemodel forinpatient mortalityaf Appendix Table3:Multivariatemodel zipcode for patient quartile Income Income quartile for patient zip code zipcode for patient quartile Income Other Risk Factors Factors Risk Other measures Comorbidity Annual Procedure Volume Factors Risk Other measures Comorbidity AHRQ Annual Procedure volume Emergent admissions only only admissions Emergent Author Manuscript only admissions Elective

Age (per 1 year increase) 1 Age (per g pr1ya nrae . 10-.3 0.02 1.00-1.03 1.0 year increase) 1 Age (per Congestive heart failure Congestive heart ogsiehatfiue15 .925 0.13 0.89-2.59 1.5 failure Congestive heart 1-19 vs.≥20TIPS/yea 1-19 vs.≥20TIPS/yea Alcoholic liver disease looi ie ies . 08-.1 0.55 0.80-1.51 1.1 Alcoholic liver disease Variceal Bleeding $64,000 or more Reference Reference more $64,000 or This article isprotected by copyright. All rights reserved. aielBedn . 14-.7 <0.01 1.44-2.97 2.1 Variceal Bleeding $48,000-63,999 $38,000-47,999 6,0 rmr eeec Reference more $64,000 or $48,000-63,999 $38,000-47,999 Coagulopathy oglpty18 .924 <0.01 1.29-2.45 1.8 Coagulopathy Renal failure ea alr . 11-.4 0.01 1.19-2.64 1.8 Renal failure $1-37,999 $1-37,999 ibts06 .314 0.22 0.23-1.41 0.6 Diabetes neto . 35-75 <0.01 3.50-17.54 7.8 Infection ibts08 .410 0.13 0.54-1.08 0.8 Diabetes Infection 3.5 2.63-4.77 2.63-4.77 3.5 Infection r r . 02-.7 0.50 0.23-2.07 0.7 . 03-.3 0.93 0.43 0.35-3.13 0.21-1.95 1.1 0.6 OR OR . 09-.1 0.13 0.94-1.01 1.0 . 01-.5 0.13 0.10-1.35 0.4 . 18-.2 <0.01 1.89-9.52 4.2 . 01-.6 0.91 0.15-5.26 0.9 . 03-.8 0.50 0.30-1.78 0.7 . 11-.1 0.03 1.10-6.41 2.7 . 20-01 <0.01 2.00-10.10 4.5 OR OR . 11-.5 0.01 1.11-2.15 1.5 . 05-.5 0.60 0.59-1.35 0.9 . 03-.8 0.01 <0.01 0.35-0.88 0.33-0.80 0.6 0.5

Hepatology Hepatology

95% CI 95%CI 95% CI 95%CI ter emergentadmissionsforTIPS. ter elective admissions forTIPS. p value p p value p <0.01 c-statistics c-statistics 0.895 0.895 0.738 0.738

Page 28of